by Robert Solow
Are you afraid of asteroids hitting earth? No? Then why would you worry about the end of work?
The belief and fear that automation and mechanization are about to make most human labor obsolete, and thus create unavoidable mass unemployment, seems always to be simmering below the surface, like a passive volcano. It erupts, predictably, every time our economy goes through a period of prolonged, high unemployment. When I started to work at the Council of Economic Advisers in the Kennedy administration more than 50 years ago, one of my first tasks was to look into the credibility of just such a belief, couched in language very much like what we hear now. It seemed to me that there was very little behind the scary rhetoric. The catastrophe didn't happen then, and it is extremely unlikely to happen now.
One reason is that there are approximately as many reasons to expect the pace of economically effective technological change to slow down as to accelerate. Another reason is that even the advanced economies of the world seem to be some distance from satiation with goods and services. And most of the world's people live at a level of income and consumption far below what the advanced economies have already achieved. Even a poorly managed world economy should welcome a large increase in productivity given the good use that could be made of a large increase in total output. (The environmental and natural resource consequences of such a development are a truly serious issue, especially in a poorly managed world economy. I am ignoring them now only because they are not what the fear-of-automation question is about.)
I hesitate to mention it, because it would require more competent economic management than we get, but most workers would be glad to have more leisure if their material standard of living were to improve simultaneously. Dramatic increase in productivity, from automation and robots, would make that possible.
A Downward Spiral
So the fear of automation is rather like the fear of collision with an enormous asteroid. But it is not a harmless fear. It does harm in two ways. It provides a convenient excuse for those who are unwilling to face up to the unemployment problem we already have and which has little to do with technology. And it diverts attention from two more interesting questions that do arise from the progress of automation, and need to be thought about.
The first of these has to do not with the total number of jobs but with their composition by skill and income. There is considerable evidence that advanced economies find it relatively easy to provide high-end, high-earnings jobs and low-end, low-wage jobs. It is the jobs in the middle, requiring considerable but not extraordinary skill, and paying substantial wages, that are diminishing. The process is sometimes called "polarization." It has two bad consequences for the economy and society. There are inevitably more jobs at the bottom than the top; so many mid-level workers and their children face a downgrading of life-chances and expectations. And the economy is deprived of a stable source of consumer spending, especially on new goods and durables.
It is at least plausible that this polarization arises mainly from the nature of new technology, especially information technology. Computers can most easily replace simple and repetitive analytical skills, clerical and even managerial. Nevertheless it is quite possible that shifts in the nature of final demands for goods and services have also played a part in the polarization story. This is a recent and under-studied set of issues, both as to the causes of polarization and the further question of what, if anything, should be done by way of policy in response to it. Panic about the end of work is not helpful.
Uninformed Worries about the Robot Economy
The second question I have in mind is much more speculative. There is some evidence, though far from definitive, that market processes are allocating an increasing share of national income to income from property or capital and a smaller share to income from work. This development, like polarization, contributes to worsening income inequality. Suppose that there is such a trend, and that it continues. Whether it has technological roots or other sources in the economy and society, the consequence for society at large could be on the ugly side, not to mention destabilizing.
If so, what could be done about it? Redistributive policy is a technical possibility, but the political track record does not suggest that it is a likely outcome. An alternative worth thinking about is some kind of democratization of capital by means other than redistribution. This could come about through the accumulation of claims to real capital by public social security funds, perhaps via special mutual funds. One could also imagine the creation of much stronger incentives to save, aimed at families with low and middle incomes. This is admittedly vague and under-researched, but it strikes me as more interesting and potentially more useful than uninformed worries about the robot economy.